How we know what we know about the climate

Physics Today has a nice article this month on the history, development and contributions of the "A-train," a formation of satellites that orbit the planet and help us observe the climate system.  For non-experts, its a gentle and interesting introduction to many of the challenges associated with trying to observe and characterize the state of something as large and dynamic as, well... the entire world.

Without satellites, what we know about climate dynamics would be limited to what we can observe from the ground, balloons and airplanes.  While detailed measurements from the surface are essential to the science, they are expensive to collect and cannot tell us what is happening in locations far away.  Satellites, on the other hand, allow us to observe many locations rapidly at a cost that is lower than it would be if we tried to make all of those measurements from the ground.

The revolution that satellites offered us, in terms of understanding the dynamics of the planet we inhabit is often under-appreciated.  One might think that humans, inhabiting so many locations around the world, would somehow possess a single collective record of the daily temperature and daily rainfall everywhere. After all, how difficult is it to step outside with a thermometer and read it? Or to leave a bucket outside and check every day how much water fell into it?  It isn't terribly hard, but without the right institutional structure to provide funding, incentives to the observers and safe repositories for the data, it doesn't happen consistently.  A quick look at current weather stations in Central Africa compared to the Eastern Unite States compared to the Central Pacific confirms that our ground observations around the world are not of uniform density (or quality).

Here's an example of the challenges we face.  I study how changes in the environment, particularly the atmosphere, influence societies.  One question I've looked at is whether rainfall matters to the economies of the Caribbean and Central America.  To answer this question, what I'd like to do is to compare how economic outcomes (like profits in agriculture) respond to changes in rainfall. To do this, I tried to look at annual average rainfall over different countries over time.  But it turns out that getting this kind of data isn't trivial.

There are thirty-one countries south of Mexico and north of Colombia (each one is given a number on the y-axis in the figure at right).  But only about half of those countries regularly collected any rainfall data between 1950-1980.  In the figure, an open blue circle indicates that the country listed on the y-axis has at least one continuous record of rainfall for that year.  Not too many countries satisfy this criteria.  If I'm willing to work with data that's been processed a bit more, I could access statistically interpolated data that tries to use the existing station data to estimate rainfall in nearby countries.  Countries with complete records of interpolated data are given orange dots in the figure.  Notice that there's more estimated rainfall data, especially before the eighties when enough stations were collecting data to make useful estimates for the countries between stations.  But what happened between 1980 and 2000?  Why did the stations stop collecting data? And even worse, once enough stations stopped recording rainfall, the interpolated estimates become garbage, so the orange dots start to disappear too (note: the orange data stop in 2000 because of the data set, not because of data availability).  It would take a fair amount of detective work to figure out exactly why specific observations were missing, but a lot of social and political changes that might have contributed to the data scarcity were occurring in the 80's and 90's. For example, many of the countries in the region gained independence from European colonizers during that period.  If new young governments were less worried about maintaining unbroken records of rainfall than older colonial governments (which is often the case, and sometimes for good reason) than this could contribute to the breakdown of our station records.

Luckily, the early eighties were also the period when satellites first began taking regular observations of the atmosphere. And while satellite observations are never identical to surface observations, they are (1) often good enough and (2) always better than nothing.  Making the best of what we have,  gauge data, satellite data and numerical simulations can be combined to create longer, continuous and more consistent records of the environment.  The yellow dots above indicate that estimates of rainfall can be constructed using this combination of data sources.

We certainly don't know everything about these planetary scale systems, but sometimes its worth reflecting on the fact that we know anything at all.  Personally, I can't decide whether its more remarkable that someone organizes thousands of scientists around the world so that they all leave buckets outside to measure rainfall or that someone realized thirty years ago that thirty years later we would be so grateful to them for launching machines into space and leaving them there just watching for the weather.

No comments:

Post a Comment